scripod.com

THIS WEEK IN AI: GPT Astra, OpenAI vs Cursor, The Truth About Data Centers

Shownote

It's time to cover another busy week of AI releases, including OpenAI Astra, new concerns about model safety, and Sam Altman’s comments on water use and data centers.  We also discuss model updates from Meta, Google, Anthropic, and World Labs’ new system ...

Highlights

A fast-moving tour of the week’s AI releases, safety concerns, emerging world models, infrastructure debates, and hardware surprises.
00:08
OpenAI Astra may already be within reach
02:44
Less readable reasoning raises safety concerns
05:19
Cursor’s OpenAI traffic was only 5%
10:44
A single image becomes a navigable world
17:29
One prompt became a working game
21:32
The mystery device keeps appearing in public
25:04
A cybersecurity model without safeguards
27:58
Energy is the problem to solve.
32:22
An overengineered but beautiful device
35:03
Apple begins a new hardware era

Chapters

GPT Astra Release
00:00
GPT-6 and Training Breakthrough
00:59
Cursor Breakup Drama
04:07
Meta and Google Models
07:36
World Models Unleashed
10:00
Anthropic’s New Upgrades
15:23
Weekly Model Rundown
18:18
OpenAI Device Mystery
19:15
Rogue Cybersecurity Model
23:51
Data Centers Bring Revenue
25:47
CyberCabs Hit Austin
28:05
Dyson’s AI Toothbrush
30:38
Apple’s New Hardware Era
33:06
Week in Review
35:05

Transcript

Ejaaz: OpenAI Astra, also known as GPT-6, is on the verge of being released. We've seen a bunch of updates from the OpenAI side. If you're listening to this episode right now, you may have access to this. This is OpenAI's most powerful model yet, and it's ...